close
Warning:
Can't synchronize with repository "(default)" (/project/cmucl/public_html/git/cmucl.git does not appear to be a Git repository.). Look in the Trac log for more information.
 Timestamp:

10/04/14 21:18:33 (3 years ago)
 Author:

rtoy
 Comment:

Fix up markup.
Legend:
 Unmodified
 Added
 Removed
 Modified

v1

v2


26  26  * Accuracy of trig functions improved. (Previously, they could be 
27  27  off by one or two ulp.) 
28   * RUNPROGRAM accepts :ELEMENTTYPE to allow specifying the 
29   element type for streams that are created for RUNPROGRAM. (See 
 28  * {{{RUNPROGRAM}}} accepts {{{:ELEMENTTYPE}}} to allow specifying the 
 29  element type for streams that are created for {{{RUNPROGRAM}}}. (See 
30  30  ticket:92) 
31  31  
32  32  * Changes 
33  33  * Update to ASDF 3.1.3. 
34   * When *PRINTCASE* is :DOWNCASE, integers are printed with 
 34  * When {{{*PRINTCASE*}}} is {{{:DOWNCASE}}}, integers are printed with 
35  35  lowercase letters when needed. 
36   * Microoptimize KERNEL:DOUBLEFLOATBITS for x86/sse2. 
37   * Add microoptimization for unary FTRUNCATE for x86/sse2. 
38   * Compiler can derive the types of the results of DECODEFLOAT. 
39   * Main disassembly routines in the DISASSEMBLE package accept 
40   :BASE, :CASE, and :RADIX keyword arguments, like 
41   DISASSEMBLE:DISASSEMBLE. 
 36  * Microoptimize {{{KERNEL:DOUBLEFLOATBITS}}} for x86/sse2. 
 37  * Add microoptimization for unary {{{FTRUNCATE}}} for x86/sse2. 
 38  * Compiler can derive the types of the results of {{{DECODEFLOAT}}}. 
 39  * Main disassembly routines in the {{{DISASSEMBLE}}} package accept 
 40  {{{:BASE}}}, {{{:CASE}}}, and {{{:RADIX}}} keyword arguments, like 
 41  {{{DISASSEMBLE:DISASSEMBLE}}}. 
42  42  * Micro optimizations for sse2 to allow descriptor regs and stack 
43  43  tns for many operations so that the operands can be in memory 
… 
… 

45  45  instruction. 
46  46  * Support for darwin/ppc 10.5 and gcc 4.8. 
47   * Microoptimization: (expt 1 power) just returns 1 or 1 
 47  * Microoptimization: {{{(expt 1 power)}}} just returns 1 or 1 
48  48  without actually computing the power. 
49  49  * Runtime checks for zero pages on newly allocated pages has been 
… 
… 

58  58  to fdlibm instead of implementing them in Lisp. 
59  59  * Source location information has been added for 
60   DEFINECONDITION. (From Helmut Eller.) 
 60  {{{DEFINECONDITION}}}. (From Helmut Eller.) 
61  61  * The lisp executable is now compiled to use SSE2 on x86 machines; 
62  62  CMUCL will not run on chips without SSE2 anymore. 
63   * (cosh 1000d0) signals an overflow error as it 
 63  * {{{(cosh 1000d0)}}} signals an overflow error as it 
64  64  should. Previously, it just incorrectly returned infinity. 
65   * Deprecating FLOATTRAPPINGNANP in favor of 
66   FLOATSIGNALINGNANP. These kinds of numbers are called 
 65  * Deprecating {{{FLOATTRAPPINGNANP}}} in favor of 
 66  {{{FLOATSIGNALINGNANP}}}. These kinds of numbers are called 
67  67  signaling NaNs, not trapping NaNs. 
68  68  * Changed the printer to print "Signaling" instead of "Trapping" 
… 
… 

78  78  failing if cmucl is unable to interpret the instruction that 
79  79  caused the exception. 
80   * Microoptimization for sparc in rounding a doublefloat to a 
 80  * Microoptimization for sparc in rounding a {{{DOUBLEFLOAT}}} to a 
81  81  32bit signed integer. 
82   * The compiler can now derive the type of ROUND. (But more work is 
 82  * The compiler can now derive the type of {{{ROUND}}}. (But more work is 
83  83  needed.) 
84   * Fix an issue with multiprocessing destroyprocess doesn't take 
 84  * Fix an issue with multiprocessing {{{DESTROYPROCESS}}} doesn't take 
85  85  effect for a schedulable process because it has no runreason 
86  86  and/or arrestreason. 
… 
… 

90  90  hyperbolic functions have been updated. Instead of following 
91  91  the text in the CLHS, the values are defined according to the 
92   definitions in the formulas. In particular (atanh x) for x real 
 92  definitions in the formulas. In particular {{{(atanh x)}}} for x real 
93  93  and on the branch cut is now continuous with different 
94  94  quadrants. This differs from the description of the branch cut 
95   for atanh in the CLHS. 
96   * CLEAROUTPUT was not actually doing anything. Now, CLEAROUTPUT 
 95  for {{{atanh}}} in the CLHS. 
 96  * {{{CLEAROUTPUT}}} was not actually doing anything. Now, {{{CLEAROUTPUT}}} 
97  97  will discard any buffered data that has not been written out 
98  98  yet. 
… 
… 

108  108  * The float constants for various limits are defined for 
109  109  doubledouble floats such as 
110   leastpositivedoubledoublefloat. 
 110  {{{LEASTPOSITIVEDOUBLEDOUBLEFLOAT}}}. 
111  111  * When reading very small float values, round to 
112  112  leastpositivefoofloat if possible. 
113   * (log 0w0) and (log 0w0) returns values analogous to the 
 113  * {{{(log 0w0)}}} and {{{(log 0w0)}}} returns values analogous to the 
114  114  doublefloat versions. 
115   * Fix bug in printing MOSTNEGATIVEFIXNUM. 
116   * For Gray streams, CLEAROUTPUT was calling STREAMFORCEOUTPUT 
117   instead of STREAMCLEAROUTPUT, so the output wasn't actually 
 115  * Fix bug in printing {{{MOSTNEGATIVEFIXNUM}}}. 
 116  * For Gray streams, {{{CLEAROUTPUT}}} was calling {{{STREAMFORCEOUTPUT}}} 
 117  instead of {{{STREAMCLEAROUTPUT}}}, so the output wasn't actually 
118  118  cleared. 
119   * In some situations KERNEL:DOUBLEFLOATBITS on x86 would cause a 
 119  * In some situations {{{KERNEL:DOUBLEFLOATBITS}}} on x86 would cause a 
120  120  segfault. This has been fixed. 
121  121  * For Linux, motifd is no longer a 64bit app. 
122   * (exp 1d0) now returns the correctly rounded value of 
 122  * {{{(exp 1d0)}}} now returns the correctly rounded value of 
123  123  e. Previously, it was off by one bit. 
124   * INTEGERDECODEFLOAT returns the correct values for denormal 
125   doubles. As a sideeffect of this fix, DECODEFLOAT returns the 
126   correct values for denormals, and SCALEFLOAT scales denormals 
 124  * {{{INTEGERDECODEFLOAT}}} returns the correct values for denormal 
 125  doubles. As a sideeffect of this fix, {{{DECODEFLOAT}}} returns the 
 126  correct values for denormals, and {{{SCALEFLOAT}}} scales denormals 
127  127  correctly. 
128   * EXT:UNIXNAMESTRING no longer returns NIL if a directory does 
 128  * {{{EXT:UNIXNAMESTRING}}} no longer returns {{{NIL}}} if a directory does 
129  129  not exist. This was a regression from at least 18a. 
130  130  